Expression of Neural Stem Cell Surface Marker CD133 in Balloon Cells of Human Focal Cortical Dysplasia


Address correspondence and reprint requests to Dr. I.M. Najm at Section of Adult Epilepsy, Department of Neurology, S51, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, U.S.A. E-mail:


Summary: Purpose: Focal cortical dysplasia (CD) is characterized by the presence of dysmorphic neurons, laminar and columnar disorganization. A few patients with CD have balloon cells intermixed with dysmorphic neurons. The cellular characteristics of balloon cells remain unknown. This study was intended to determine further the cellular characteristics of balloon cells.

Methods: Neocortical tissue resected from five patients with medically intractable focal epilepsy due to CD was studied. The presence of balloon cells (large opalescent cells with eccentric nuclei) was confirmed in all five patients by using cresylecht violet staining. Immunocytochemistry used antibodies against markers of pluripotential stem cells (CD133), multipotential progenitor cells (nestin), antiapoptotic gene products (Bcl-2), immature neurons (β-tubulin 3, TUJ1), immature glia (vimentin), mature neurons (MAP2 and NeuN), and astrocytes (glial fibrillary acidic protein; GFAP).

Results: Balloon cells (BCs) were found to be immunoreactive to Bcl-2 (46%), vimentin (41%), Nestin (28%), CD133 (28%), MAP2 (27%), GFAP (14%), and TUJ1 (10%). An extremely small number of BCs were immunopositive for NeuN. Confocal double labeling showed that balloon cells were dually immunopositive for CD133/nestin; CD133/GFAP; CD133/Bcl-2, and nestin/GFAP.

Conclusions: These results show that balloon cells are heterogeneous cell populations expressing cell-surface markers for pluripotential stem cells and proteins for multipotent progenitors, or immature neurons/glia. The presence of stem cell/progenitor markers in the balloon cells could be due to a persistent postnatal neurogenesis or early embryonic insult that resulted in arrest of proliferation/differentiation at their early stages. Additionally, the coexpression of Bcl-2 in CD133-positive balloon cells suggests that a resistance to programmed cell death may be involved in the pathogenesis of cortical dysplasia.

Focal cortical dysplasia (CD) are increasingly recognized as a frequent cause of drug-resistant neocortical epilepsy that is amenable to surgical resection (1,2). Microscopically, CDs display a broad spectrum of structural changes that range from disorganized cortical lamination and orientation to aberrant neuronal cytoarchitecture (3–5). Among the morphologic abnormalities of CDs is the presence of balloon cells (BCs) that are characterized by eccentric, pleomorphic nuclei and large, homogeneous, eosinophilic cytoplasm (3–5). These BCs have a propensity to localize outside of the temporal lobes and to cluster in the deep cortical layers and at the junction between gray and white matter. It was previously hypothesized that such structural abnormalities are the result of aberrant patterns of proliferation, differentiation, migration, maturation, and/or apoptosis of neuronal precursors and neurons during cortical development (6–9).

The origin, pathogenesis, and role of the balloon cells in CD and epileptogenesis remain to be elucidated. The histologic appearance and the location of BCs suggest that they are developmentally immature. Furthermore, by using a silver-impregnation technique, Derosa et al. (10) showed the presence of nucleolar organizer regions that were believed to be involved in cell proliferation in the BCs. No reports have been published on the presence of BCs in nonepileptic and otherwise neurologically normal brain autopsy tissue in children.

Critical steps in uncovering these mechanisms are to identify developmental characteristics and the state of maturity of BCs. Over the past decade, several studies have evaluated the cellular expression of intermediate filament proteins in the BCs. Nestin, vimentin, and glial fibrillary acidic protein (GFAP) belong to the intermediate filament families: nestin, may act as a marker of neural precursors (11); vimentin is expressed in less mature glial cells (12); and GFAP labels not only astrocytes but also their precursors (13). Several authors had reported the expression of these intermediate filament proteins in the balloon cells (8,14–17). Furthermore, transcription of genes for nestin and vimentin was enhanced in the BCs (18). Over the past several years, much more attention has been focused on nestin. Because initial studies have shown that nestin expression occurs early during central nervous system (CNS) development (19), subsequently nestin was considered a marker for neural progenitors (20–23).

Recently, human hematopoietic stem cell antigen AC133, also called CD133, was shown to be expressed in both hematopoietic and neuropoietic cells (24). CD133-positive cells are capable of neurosphere initiation, self-renewal, and multilineage differentiation at the single-cell level (24). Therefore CD133 has been used to identify neural stem cells in the human brain (24–26).

In an attempt to define the developmental identity of BCs, we studied the expression of CD133 in BCs and its coexpression with other intermediate filament proteins. We hypothesized that BCs are a heterogeneous population with characteristics of neural stem cells, immature neurons/glia, or mature neurons/astrocytes.



The focal CD samples containing BCs used in this study were obtained from five patients who were operated on for the treatment of drug-resistant epilepsy (Table 1). All patients were evaluated at the Cleveland Clinic Epilepsy Center, as previously described (27,28). The presurgical evaluation included scalp continuous video-EEG recordings, high-resolution magnetic resonance imaging (MRI), fluorodeoxyglucose (FDG)-positron emission tomography (PET), and neuropsychological studies. This study was approved by the Institutional Review Board of the Cleveland Clinic Foundation.

Table 1. Patient demographics and electroclinical data


Age at
surgery (yr)

Age at
seizure onset




  1. AEDs, Antiepileptic drugs.

1 16 moMCarbamazepine (Carbatrol); topiramate (Topamax); phenobarbitalLeft hemisphereLeft hemisphere (especially frontal)Left hemisphere
2 49 moFClobazam (Frisium); lamotrigine (Lamictal)Right frontalRight frontal poleRight frontal
3104 yrFPhenytoin (Dilantin)Right frontalRight mesial frontalRight frontal
4122 yrFDivalproex (Depakote)Right frontalRight orbitofrontalRight frontal
5 71 yrMDivalproex (Depakote); phenytoin (Dilantin); phenobarbitalRight frontalMesial right frontoparietalRight frontoparietal convexity

Tissue preparation

Four patients had partial lobectomies, and one patient underwent a hemispherectomy. As previously described (28,29), representative portions of the specimens were received fresh from the clinical neuropathologist in the operating room, cut into blocks 0.5–0.7 cm thick, and immersion-fixed in 4% paraformaldehyde at 4°C for 48 h. The fixed tissue blocks were cryoprotected overnight in 20% buffered sucrose and then were frozen quickly in crushed dry ice and cut into 30-μm sections on a cryostat (model 1850 CM; Leica, Heidelberg, Germany). Immediately adjacent sections were processed simultaneously for immunocytochemistry (ICC) with antibodies to CD133 [1μg/ml; mouse immunoglobulin G (IgG); Miltenyi Biotec, Bergisch Gladbach, Germany]; Bcl-2 (1.0 μg/ml; rabbit IgG; Santa Cruz Biochemicals, Santa Cruz, CA, U.S.A.); nestin (1:300; rabbit serum, Chemicon, Temecula, CA, U.S.A.); MAP2 (0.5 μg/ml; mouse IgG; Boehringer Mannheim, Germany); GFAP (0.5 μg/ml; mouse IgG; Boehringer Mannheim); NeuN (1μg/ml; mouse IgG; Chemicon), TUJ1 (2.5 μg/ml; mouse IgG; Covance, Richmond, CA, U.S.A.); vimentin (1:300; mouse IgG; Pharmingen Carpinteria, CA, U.S.A.). t-Butyldimethylsilyl (TBS; 0.05 M, pH 7.6) was used as the rinsing buffer throughout the ICC staining procedure on free-floating sections with rinses between each step: 5 min in 3% hydrogen peroxide/10% methanol in TBS; 60 min in a blocking solution of 1.5% normal horse serum in TBS; 18 h overnight in primary antibody diluted in TBS containing 1% normal horse serum; 35 min in biotinylated horse anti-mouse IgG antibody; 60 min in avidin DH and biotinylated horseradish peroxidase H reagents. To visualize the immunoreactivity, we reacted the sections for 7 min in 0.05% 3,3′-diaminobenzidine tetrahydrochloride and 0.01% H2O2 in TBS. The reaction was terminated through transfer of the sections into ice-cold TBS. The tissue sections were then mounted on slides, air-dried, and coverslipped. Control experiments for ICC were performed by omission of the primary antibody by using the same staining protocol; no specific/cellular immunocytochemical staining was seen in the absence of specific primary antibodies. All the antibodies that were used in these experiments were purchased from various commercial sources mentioned earlier. Based on the statements provided by the producers, all the antibodies used were reported to be specific for each target protein, and no cross-reactivity was observed.

Double labeling and confocal microscopy

For double-labeling experiments, sections were first incubated in a blocking solution containing 5% normal donkey serum in TBS for 1 h, and then simultaneously incubated in two primary antibodies raised in two different species at 4°C for 3 days (CD133 of mouse IgG with nestin of rabbit IgG; CD133 of mouse IgG with Bcl-2 of rabbit IgG; CD133 of mouse IgG with GFAP of rabbit IgG; 3 μg/ml; Promega; nestin of rabbit IgG with GFAP of mouse IgG; Boehringer Mannheim (CD133 with MAP2 of rabbit serum, 1:300; Chemicon). Sections were washed in TBS and incubated in fluorescent secondary antibodies. Antibodies to CD133 and GFAP of mouse IgG were visualized by using donkey anti-mouse antiserum labeled with fluorescein isothiocyanate (FITC; Jackson Laboratories, West Grove, PA, U.S.A.). Antibodies to nestin, Bcl-2, GFAP, and MAP2 were visualized by donkey anti-rabbit antiserum coupled to Texas red (Jackson Laboratories, West Grove, PA, U.S.A.). FITC and Texas red–conjugated secondary antibodies (diluted at 1:200) were applied for 1 h. Sections were then rinsed with TBS and mounted in a Mowiol-based (Calbiochem) mounting medium containing 0.1%para-phenylenediamine hydrochloride. The control experiments consisted of omission of the primary antibodies. No specific staining was seen in the absence of any one of the primary antibodies. Fluorescent-stained sections were examined by using a Leica TCS-NT confocal laser scanning microscope (Leica Lasertechnik GmbH, Heidelberg, Germany), and the green and red channels were simultaneously acquired with the ×20 (oil) lens.


As shown in Fig. 1 and previously reported (3–5), CD was histopathologically defined by the presence of laminar and columnar disorganizations, lacking both the horizontal layering and vertical orientation. BCs were found in all five patients included in the study. All BCs were found primarily in the white matter, some patients having higher number of BCs than others (no direct correlations were found between the BC densities and either age at seizure onset or age at surgery). Balloon cells were frequently located in the gray and white matter junction and extended into the white matter (Fig. 1B), which resulted in the blurring of the white and gray matter junction when compared with normal cortex (Fig. 1B). Dysplastic cortex was at times bordered by normal-appearing cortical areas with well-preserved laminations and unipolar orientation of apical dendrites toward the pial surface, with normal definition of the gray/white matter junction (Figs. 1A and C). In the cresylecht violet (CV) staining, the balloon cells showed the typical features of eccentric nuclei with large pale cytoplasm (Fig. 1D).

Figure 1.

Photomicrograph of cresylecht violet staining from patient 2, showing a relatively normal-appearing cortical area (A, C) and the dysplastic area containing balloon cells (B, D). In the dysplastic area (B), both vertical and horizontal laminations are disrupted, and the junction between the gray and white matter is blurred because of the presence of dysplastic neurons in the subcortical white matter. The balloon cells are distributed mainly in the subcortical white matter and extend into deep white matter. At the higher magnification, the balloon cells are characterized by strikingly large opalescent cytoplasm with eccentric nuclei (D). By contrast, the normal-appearing cortical area shows well-laminated cortical pyramidal cells (A) with their dendrites positioned toward the pial surface (C). Scale bars: 400 μm in A and B; 100 μm in C and D.

ICC using antibodies to CD133 showed no cellular labeling in the regions that were devoid of BCs. In BC-containing regions, cells showing immunoreactivity to CD133 were mainly found in the deep cortical layers and at the gray–white matter junction (Fig. 2B). At higher magnifications, these CD133 immunoreactive cells were identified as BCs (Fig. 3A). Moreover, some BCs were immunoreactive to Bcl-2 (Figs. 2C and 3H). Intense immunoreactivities to antibodies against intermediate filament proteins GFAP, nestin, vimentin, and TUJ1 also were detected in some BCs (Fig. 3B, C, E, and G). Antibodies against mature neuronal markers MAP2 and neuronal nuclear (NeuN) also stained a much smaller population of balloon cells (Fig. 3F and H). On quantification of the BCs expressing various proteins and as shown in Fig. 3 (bar graph), Bcl-2 and vimentin were the most commonly found in BCs (46% and 41.5%, respectively). Other expressed markers in the BCs included: CD133 (28.3%), nestin (27.7%), and MAP2 (27%). A smaller population of BCs expressed GFAP (14.2%) and TUJ1 (10.3%), and the expression of NeuN was almost nil among BCs (0.2%).

Figure 2.

The balloon cell–containing area from patient 2. Adjacent sections were processed with cresylecht violet staining (A) or immunocytochemically stained with CD133 (B) and Bcl-2 (C). The balloon cells are present in the subcortical white matter and extend into the deep white matter (A), and they are immunoreactive to CD133 (B) and Bcl-2 (C). Scale bar: 100 μm.

Figure 3.

Immunocytochemical characterization of balloon cells. Balloon cells that are identified with eccentric nuclei are immunoreactive to protein markers for CD133 (A), glial fibrillary acidic protein (B), nestin (C), NeuN (D), vimentin (E), MAP-2 (F), TUJ1 (G), and Bcl-2 (H). Scale bar: 100 μm.

Further to characterize the BCs, the colocalization of antibodies against CD133 with antibodies to intermediate filament proteins (nestin and GFAP) were studied by using confocal microscopic analysis. Coexpression of CD133 with either nestin or GFAP was clearly detected in the BCs (Fig. 4A and D). The double expression of CD133 and Bcl-2 was confirmed in a subset of BCs (Fig. 4C). A population of BCs was shown to be immunoreative to both nestin and GFAP (Fig. 4B). Double staining for CD133 and MAP2 failed to reveal any colocalization of these proteins in the BCs (data not shown).

Figure 4.

Confocal double-immunostaining images of balloon cells. A: CD133 (green), nestin (red), and coexpression of CD133 and nestin in balloon cell (yellow, arrow). B: Glial fibrillary acidic protein (GFAP; green), nestin (red), coexpresion of GFAP and nestin in balloon cells (yellow, arrows). C: CD133 (green), Bcl-2 (red), and coexpression of CD133 and Bcl-2 in balloon cell (yellow, arrow). D: CD133 (green), GFAP (red), and coexpression of CD133 and GFAP (yellow, arrow). Scale bar: 100 μm.

MAP2 and GFAP, the neuronal and glial cytoskeletal proteins, respectively, have been previously used as markers for the definition of the neuronal or glial phenotype of BCs (30). In agreement with this report as well as another (31), we observed that some BCs were immunoreactive to MAP2, whereas others were reactive to GFAP. We did not observe any colocalization of these two proteins in the balloon cells (data not shown).


The presence of BCs in the context of other stigmata of CD that include cortical dyslamination intermixed with dysmorphic neurons is a subtype of focal CD (4,5). Recent studies have suggested that the occurrence of BCs is the consequence of a failure of neural proliferation, differentiation, or migration (32–36). In the present study, we showed positive immunoreactivity of CD133 in the BCs. The coexpression of CD133 with nestin and GFAP in BCs was confirmed with confocal microscopy.

CD133 is a glycoprotein containing five transmembrane domains. It has been used to isolate human hematopoietic stem cells (37). Uchida et al. (24) showed that this human hematopoietic stem-cell protein CD133 also is expressed in neuropoietic cells of the human fetal nervous system. This fact makes CD133 a unique cell-surface marker for the detection of brain stem cells. Most important, along the continuum of neurogenesis from very primitive stem cells (pluripotent stem cells) to more restricted stem/progenitors (multipotent stem cells) to fully differentiated cells, CD133 labels very early pluripotent neural stem cells. A characteristic feature of this protein is its rapid downregulation during cell differentiation (38–41).

The presence of CD133 protein in the BCs is interesting, as it may be due to one of two hypothetical possibilities:

  • 1The expression of this protein in the BCs may suggest that these cells fail to mature fully and therefore continue to express embryonic genes and proteins. Subsequently, these immature cells may lack some of the cellular machinery for migration. This may explain at least in part the localization of the BC in the white matter and the gray/white matter junction (deep cortical layers). Hence, we could hypothesize that the presence of BCs may signify that the intrauterine insults occurred at earlier stage of embryonic development as compared with other pathologic types of CD.
  • 2CD133-positive BCs may represent a population of neural stem cells. However, to confirm this interesting possibility, it must be proven in vitro that BCs are capable of self-renewal and can give rise to cells other than themselves though asymmetrical cell division. Long assumed to be exclusive to the developing brain, neural stem cells have now been shown to persist in postnatal and adult human brain. The explant culture studies on the adult human periventricular subependymal zone (SEZ) and mature human hippocampus have demonstrated a de novo neurogenesis (42,43). It also has been shown that postnatal human SEZ and hippocampus (up to 57 years old) contain pluripotent stem ancestors. These later cells have been isolated and cultured in vitro to produce the distinct geometric structures termed “neurospheres” (23). In addition to the adult human SEZ and hippocampus, adult human neocortex (up to 52 years old) has been shown to contain stem progenitors. By using in vitro cortical cultures with tritiated thymidine, Pincus et al. (44) demonstrated that stem cells in the adult human neocortical explants were capable of mitotic neurogenesis. Furthermore, Palmer et al. (42) proved that stem cells that were obtained from adult temporal neocortex indeed produced three lineages of cells: neurons, astrocytes and oligodendrocytes. Most interestingly, in the study by Pincus (43), all the neocortical tissues were surgically obtained from patients with medically refractory epilepsy. These studies in conjunction with our observation raise an interesting notion that certain epileptic cortices may harbor residual neural stem cells. However, the pathogenesis of these immature cells in some types of epilepsies remains to be elucidated. One theory is that certain inciting events during the prenatal/in utero period affect the neural stem cells in the ventricular zone area and result in abnormal dynamic processes including proliferation, maturation, migration, and terminal differentiation of neural stem cells. An alternative hypothesis is that chronic seizures cause certain genetic changes that may be involved in the facilitation of postnatal neurogenesis.

The detection of antiapoptotic gene products, such as Bcl-2 in the BCs and its colocalization with CD133, raises another interesting hypothesis regarding the pathogenesis of BCs. Putative apoptotic cell death occurs within the SEZ (43,44). Bcl-2 protein is involved, associated with neuronal immaturity (45). Recent studies suggest that Bcl-2 may act as a protective signaling molecule against cell death, either through an “antiapoptotic” activity (46), or as a “survival-promoting protein” (47). Our current findings that BCs are located outside the SEZ support the notion that Bcl-2 may facilitate the survival of BCs by suppressing the programmed cell death.

Nestin is a class VI intermediate filament protein. Unlike that of CD133, nestin expression is associated with a wide spectrum of cells during CNS development. For example, nestin expression is initiated in the neuroectoderm and subsequently in the multipotential neuroepithelial stem cells (11). In addition to its expression in the undifferentiated neural stem/progenitor cells, nestin also is expressed in the transitional progenitor cells that have already begun a differentiated pathway along a neuronal or glial lineage (22). In the present study, CD133 and nestin were coexpressed in a subpopulation of BCs. These data suggest that some CD133-positive BCs are in the transitional stage from the primitive pluripotent stem cells toward multipotent progenitor cells.

Additionally, we found a colocalization of CD133 in a subset of GFAP-immunoreactive BCs. This finding implies that certain GFAP-positive cells might exhibit characteristics of stem cells; such a notion is supported by previous studies (48,49). In agreement with previous reports (8,50–54), we found that BCs are immunoreactive to vimentin and TUJ1, which are generally considered to be markers for less mature glia and neuronal lineage, respectively. Similarly, it is our impression that fewer BCs are labeled by mature neuronal markers such as MAP2 and NeuN. In addition, our confocal studies failed to detect any colocalization of CD133 with MAP2. These findings lend further support to the hypothesis that BCs failed to express their cellular commitment during development (4,8,54).

In conclusion, we found BCs expressing neural stem cell surface marker CD133. Whether CD133-positive BCs are stem cells remains speculative, because this was an observational study, and it is not possible to investigate the spatiotemporal development of the BCs. It is possible that CD133-positive BCs are not functioning stem/progenitor cell; rather they express only some of the embryonic proteins. However, to define BCs as neural stem cells, further studies are needed to prove that BCs are capable of proliferation in an undifferentiated but pluripotent state and have the ability to differentiate into more specialized adult/differentiated/functional cells.


Acknowledgment:  This work was supported by National Institute of Health Grants K08 NS-02046 and R21 NS42354 to Imad M. Najm. We thank all members of the Cleveland Clinic Epilepsy Center for their collaboration during the preoperative phase of the tissue characterization. The research on resected human brain was approved by the Cleveland Clinic Foundation Institute Review Board.